Shared Custody Arrangements:
Pilot Interviews With Parents

2004-FCY-5E

3. FEASIBILIY OF FUTURE RESEARCH

One goal of this undertaking was to determine the feasibility of conducting a national study on custody arrangements. Any Canada-wide study of shared custody or other types of custody arrangements would require a significant commitment of resources. In order to make such a commitment, the Department of Justice Canada required information on the feasibility of conducting such a study, and some idea of an appropriate methodology to utilize. This small-scale pilot has been able to provide useful information in several areas including:

  1. The ability to obtain a sample of parents and the methodology to use;
  2. The logistical aspects of conducting research involving interviews with divorced and separated parents;
  3. The potential feasibility of interviewing children; and,
  4. Challenges with respect to the reliability of information obtained through interviews with parents.

A discussion of these issues is contained in the following sections.

3.1 Case Identification

The primary challenge of conducting a national study, from a logistical perspective, will be to obtain a sufficiently large sample of parents. Such a sample would need to include cases of each targeted type of custody arrangement (e.g. shared, sole and split), and identify enough cases on a national basis from which to draw a random sample that would be representative.

3.1.1 Sample Considerations for a National Study

In any future project on custody arrangements in Canada, it is essential to have a nationally representative sample. This sample should not only include families with shared custody arrangements, but also those with sole-mother, sole-father and split custody. The pilot gave us an indication that we could identify and contact parents with shared custody. There is no reason to believe that it would be difficult to expand the categories to include other custody arrangements.

The courts in Alberta have tended recently to issue a greater proportion of shared custody orders than some other provinces. Nationally, shared custody arrangements account for an average of 5.3 percent of all divorce custody orders. In Alberta, shared custody orders average about 5.5 percent (Department of Justice Canada, 1999). These differences mean that some provincial variation will need to be taken into account during sampling.

3.1.2 Divorced and Separated Families

The pilot sample included only cases in which parents were divorced. However, in the wider population, many parents never marry, and many married parents who separate, never divorce. Thus, another issue to be considered is the inclusion of families that have separated, but not divorced. If a decision is made to include separated families, the methodology of the proposed study will be affected to some degree (to allow for identification of these families). Also, consideration must be given to the fact that in these cases, the provinces have jurisdiction. Thus any decision to include separated parents would require consideration of different provincial legislative regimes.

3.1.3 National Case Identification

The methodology used in this study to identify potential cases proved to be effective for the purposes of selecting a small sample of cases in which parents divorced and were granted an order for shared custody. However, even within Alberta, the cases selected were not necessarily representative of all divorces in that province. The Survey of Child Support Awards (SCSA), which includes cases from 16 courts in all provinces and territories except Quebec and Nunavut, could be used to identify potential cases for inclusion in a national study. This database provides a convenient source for cases in selected courts, which is easily sorted by custody type. The one significant exception to the representativeness of the SCSA cases is the fact that it does not include any cases from Quebec. However, Quebec has created a survey that includes divorces, separation and common-law cases that could be used to obtain a sample of cases living in Quebec. To obtain an even greater number of potential cases (and a more representative sample), it may be possible to supplement the cases identified from the SCSA with cases drawn directly from some courts not included in the survey. However, it should be noted that such a process would be labour intensive, and the information may be difficult to access.

3.1.4 Representativeness and Over-sampling

A decision would also have to be made as to whether to attempt to pursue representativeness beyond including several types of custody arrangements. It would be more difficult to undertake stratified random sampling on a nationwide basis, but this procedure would potentially provide a more accurate picture of custody arrangements in Canada. Also, due to the small proportions of families with shared, sole-father and split custody, it would be important to over-sample families in these categories. This would improve the long-term usefulness and value of the study. Otherwise, for example, if 4% of families in Canada had shared custody arrangements (based on their divorce outcome), and the proposed random sample was 1000 cases, only 40 families would be sampled.

3.1.5 Cultural and Language Issues

We did not attempt to identify any cultural influences on shared custody or whether shared custody is a likely alternative in different Canadian communities. In designing a national study, it may be of interest to consider this aspect in asking respondents about what influenced their choice of custody arrangement and how they put their arrangement into practice. In addition, in the pilot study, we were prepared only to conduct interviews in English or French (however all interviews were conducted in English). In a larger national study, some parents will choose to respond in languages other than English or French. Those developing a national study would have to consider whether to conduct interviews in other languages or, if not, what biases this might impose (likely as a separate study).

3.1.6 Comparison samples

In a national study, it might be advisable to include a comparison sample of intact families that could be similar to the divorced/separated sample in terms of socioeconomic and demographic factors. This addition would strengthen the research design and allow for comparisons between families that have dissolved and those still intact. Although this inclusion would increase the number of cases to be included in the prospective research, the pay-off might be worth the cost.

3.2 Recruitment Procedures and Participation Rates

As expected, the major difficulty in contacting prospective respondents was that the family-court case files often did not provide up-to-date addresses, and in many cases, phone numbers were out-of-date or not available. Where only an address was provided, an effort was made using Canada 411 and other telephone search services, to obtain a phone number. Where this was successful, the respondent was usually contacted.

At the outset of this project, a decision was made to include, where possible, both parents of a former union. Further research projects will have to consider the cost and benefits of this approach, because in our experience, this decision introduced several problems:

  1. Initial calls were made only when a correct phone number for both parents was available. This placed considerable constraint on the potential participant pool.

  2. The researcher required an accurate address for both parents in order to attempt contact with either one. In the pilot study, this occurred for only 35 cases (versus 119 cases with one address available).

  3. For the majority of analyses, only one parent from each former union was included. This procedure raised some concern about the appropriateness of obtaining a full set of data for some respondents that would not be ultimately used for the full purpose of analysing responses. In this pilot, we randomly selected one parent from each of the matched sets for our parent sample (which was the sample used for the majority of the analysis). The responses from the other parent were not used except for comparing mothers’ and fathers’ responses to certain questions. The primary value of the "matched" parent responses was in examining the agreement or shared perceptions of both parents of a former union on certain issues. The matched responses were also used to identify aspects of shared custody arrangements that appeared contentious in that the two parents had widely differing points of view on parenting issues. Having responses from both parents regarding expenses also helped to provide some preliminary information on the costs of shared parenting. These were of value in assisting the Department to plan subsequent research, but may not be as important in a follow-up national study.

In planning a larger national research project, it would be worth considering obtaining a full sample of randomly selected parents, and then approaching the former partners of a sub-sample of the original respondents for limited and specific purposes only. In this case, it might be preferable to develop two questionnaires; the new one shorter and focused on particular issues for one of the "matched" parents. This approach would be supported by the fact that for most areas, parents agreed on the major issues, with small differences. Also, the cost of fully interviewing both parents in a large sample would be prohibitive, and to some extent wasteful, as there is no way to determine which parent is "more" correct. Using information from two members of the same former union would also over-emphasize the responses of certain families.

3.2.1 Ability to Contact Potential Participants and Alternative Methodologies

The other key step in obtaining an adequate sample for a national study involves contacting the cases identified for evaluation. For the current study, court files with a least one address were available for more than half of the cases initially drawn for our sample. Thus, one might expect that similar research would require an initial sample of almost twice the number of cases required to meet sampling requirements (e.g. identify 2000 cases to obtain 1000 participants). In the pilot study, however, we decided to include both parents of a former union. This requirement greatly reduced the potential number of cases for the research.

It might be feasible to consider alternative methods of contacting potential participants. Obtaining information from court files presents a problem because of the lack of contact information in some cases. One potential avenue to explore is to approach lawyers for the contact information. The court files always contain the name and address of the lawyers who represented the parents in their divorce proceedings. It would be possible, although more time-consuming, to contact these law firms and request that they contact their clients to provide information about the research, and if they are interested, either ask them to contact the research team or give the team permission to contact them. In the case of a mail-out survey, clients could be asked to consider completing the questionnaire.

3.3 Costs and benefits of the Interview Methodology

In the present study, the decision was made to directly interview parents over the phone. The benefits of using the one-on-one telephone interview methodology are numerous and include the ability to:

  • Get a higher rate of participation because of direct person-to-person contact;
  • Ensure timely responses are made that require little investment on the part of the respondent:
  • Probe and clarify responses;
  • Ensure that the respondent understands the question;
  • Make a connection with the respondent and potentially obtain a higher quality and greater amount of information; and,
  • Determine how well the interview schedule suits the sample and make changes.

The costs of this methodology include:

  • Significant time and resources must be dedicated to each participant;
  • Increased time to translate comments into quantitative data; and,
  • A requirement for a well-trained and experienced interviewer.

In further research, this methodology could prove sound. However, other possible protocols could be explored to reduce costs in a larger scale study. Alternative methodologies could include:

  1. a mail-out survey;
  2. focus groups and group testing;
  3. file reviews, in combination with other sources of information.

Each of these options (described below) would have benefits and costs, and should be thoroughly considered, in addition to maintaining the interview methodology.

3.3.1 Mail-outs

A mail-out survey would require the development of a questionnaire that could be mailed to potential participants. In general, this approach will garner a very low rate of response from parents who lead busy lives and have no incentive to fill out yet another form. However, a survey such as this one could be paired with measures designed to increase response rates, such as calls to the home in attempt to solicit participation, or limited incentives. It’s possible that any potential cost savings through a survey of this kind would be offset by the high expense of sending out the large numbers of surveys required to obtain an adequate sample size. This problem would be more significant in the smaller provinces and territories, where there may be small samples of cases to begin with, particularly for sole-father, shared and split custody cases.

3.3.2 Group Testing and Focus Groups

Group testing sessions or alternative group methodologies, such as focus testing, could be used to collect information. The benefit of this type of approach is that a larger number of respondents can be given questionnaires at the same time. It is likely that a higher participation rate could be obtained through use of this methodology and the cost of individual interviews reduced. Further, the method provides for more contact between the interviewer and the respondents, making it easier to clarify questions if required. On the other hand, getting parents to a particular location would require significant organization and planning. Incentives might be needed to encourage parents to come to the place of testing.

3.3.3 File Reviews and Supplementary Information Gathering

File reviews, although perhaps the least preferred and least informative option, could be used to supplement any other methodology chosen, depending on the information sought.

3.4 Information Collected Through Interviews

Based on the pilot study, researchers may decide to make changes to the information solicited, however it appears that the questionnaire was well designed for shared custody arrangements. There were no questions that parents (as a whole) refused to answer or appeared to misunderstand. However, any changes to the target sample for future studies (e.g. different custody arrangements and intact families) would require the interview schedule to be adapted to different family experiences. These changes would flow from research team discussions, but would likely be heavily influenced by the structure of the current instrument. Below are some items for consideration as additional questions/areas for exploration.

3.4.1 Number of Children in the Household

In the pilot study, the number of children in the household was not determined. Only information on the number of children from the prior marriage was collected. It may be important, however, to obtain this information in a future research project. For example, in examining household expenses related to parenting, the number and age of children in the household is relevant. To factor these elements into the analysis in a larger national study, several issues will need to be considered. First, researchers should inquire not only about children of the relationship with the two parents in question, but also any children from new or past relationships who may be residing in the parents’ residences. Also, depending on the depth of analysis to be undertaken with regard to expenses, the national study may need to inquire directly about expenses related to each of the children separately, so that the influence of the children’s ages can be adequately taken into account. Finally, it should be recognized that in some cases there might be no children in a parent’s household (for example because the children are older and have moved out on their own by the time of the interview). Researchers will need to consider whether to include such cases in the sample.

3.4.2 Reliability of Responses

There were no formal tests of social desirability or examinations of reliability built into the design of the interview guide for this pilot. However, the fact that we had a sub-sample of 19 matched sets of parents meant that we could compare responses from parents on questions of fact, as one way of assessing reliability. The interview included many questions that sought descriptions of the living arrangements of the family at different points in time, questions that inquired about the division of responsibilities and expenses, and other questions for which there might be some expectation of similar responses between former partners.

For these questions, there was an overall "agreement rate" of 83 percent in our study. That is, for all instances in which former partners were asked the same factual questions, 83 percent of the time the partners responded consistently with each other. This does not indicate the extent of disagreement. For example, some questions invited a range of possible responses of greater or lesser divergence, and we have not analysed the degree of divergence between the matched parents in their responses to these questions. Further analysis on a question-by-question basis may be of interest in helping to guide the structure of questions for future interviews. For our purposes, it is sufficient to say that the degree of agreement between the matched parents is an indication that the reliability of responses on sensitive issues may be at an acceptable level. However, some disagreement did exist on questions where one might have expected high agreement, such as the duration of the couple’s marriage or how long they had been separated. This indicates that, where possible, objective sources of information should be sought. Court files have information on dates of marriages and divorces and the specific terms of divorce. Independent confirmation of these details would be useful to have as a starting point for understanding custody arrangements.

Figures provided by respondents for annual expenses in different expense categories represent estimates only; we have no way of assessing their accuracy. On this point, there may be merit in providing respondents with written explanations of the types of expenses the interviewer would be asking about, prior to the interviews. This would not alleviate the risk of some misinformation being obtained, but it might reduce the risk somewhat by limiting error due to a misunderstanding of what expense information was being requested.

3.4.3 Parenting Arrangements and Divorce Proceedings

In our pilot, we noted that a number of families had been in sole-custody arrangements prior to the divorce, and had then adopted a shared custody arrangement. We don’t know what the impetus was for the change in custody arrangement, or what the impetus was for the divorce itself, but it may be that the formality of the divorce process (e.g. the involvement of legal counsel with differing ideas of a position to assume) may lead parents to reconsider their arrangements. It may also be that a resolution of the parenting arrangements is a factor in the decision to divorce. In either case, the divorce process may affect the type of custody in place and the way the details of the arrangements are worked out in individual cases. To the extent that this is true, it will be important to look at non-divorce cases to understand more fully the factors influencing custody arrangements and the outcomes for children.

3.5 Interviews with Children

It is important to assess the outcomes of custody arrangements for children. The Department of Justice Canada is interested in the possibility of obtaining some information directly from children as part of the national research project. The methods to obtain this data could range from individual interviews to focus group testing, such as was undertaken by the Federal/Provincial/Territorial Task Force on Custody and Access.

In the current study, parents were asked to comment on their children’s adjustment and satisfaction. But more importantly they were asked whether they would feel comfortable with the participation of their children in a future study. The majority of respondents (over 60 percent) indicated their willingness, with some caveats and suggested limits.

There are, of course, sensitivities and confidentiality issues associated with interviewing children on a subject as personal and potentially distressing to them as their living situation after their parents separate. But any protocol developed would consider these issues, including follow-up referrals in the event intervention is required. As well, certain methodologies (e.g. group testing with sufficient supervision) would likely be more appropriate than others.

Finally, we need to recognize the possibility that parents who are not getting along and are having disagreements about custody, access, child support or other issues relating to their children, may be more reluctant to have their children discuss their situation with an interviewer. No such linkage emerged from our small sample of parents who did not appear to be getting along together, but there were too few of them to draw any inferences.